Candidate genes were chosen for one or more of the reasons discussed in section 1.4.2.2: expression, function, and homology or relatedness to known associated genes. All genes were of known map position, and polymorphic markers were chosen that had been identified within the gene, or mapped as close to the gene as possible.
The majority of markers for genes without published intragenic markers were chosen by entering the gene name or accession number into a search for the STS mapped region of the gene on the SCIENCE 96 gene map consortium data web site, entitled Research Tools (http://www.ncbi.nlm.nih.gov/SCIENCE96/ResTools.html). The region for the gene was given with respect to the (AC)n repeats of the Genethon map (section 1.4.1.1). Details of the markers including PCR primer sequences, allele product size ranges, the number of alleles identified, and the level of heterozygosity were accessed from the Genethon web site (http://www.genethon.fr), unless otherwise referenced. The sex-averaged distances between the markers were also given, and were used to calculate approximate distances of the marker from the candidate gene. The best marker was one closest to the gene with a heterozygosity over 0.75. The informative markers presented ranged in heterozygosity from 0.48 to 0.93, although the majority were between 0.7 and 0.9. Approximate maximum distances of the markers from each gene ranged from 0 to 10 cM, although most were within 5 cM.
In the case of FGFR4, an exact map position with respect to nearby markers was not available. In order to identify a nearby marker, a radiation hybrid screen of the Genebridge 4 panel was carried out. This whole genome panel had been produced by fusing irradiated diploid human fibroblasts with recipient hamster cells (Gyapay G, 1996). A PCR reaction (using primers FGFR4F and -4R (Table 2.3)) that would amplify a 278 bp region of FGFR4 including some of exon 7 was carried out on each of the 93 cell line samples of the panel. After amplification, each reaction sample was electrophoresed on a 1.5% agarose gel, and cell lines in which the amplification had
occurred were noted. The results were submitted (http://www-genome.wi.mit.edu/cgi- bin/contig/rhmapper.pl) to map the fragment relative to the Whitehead Institute/MIT Center for Genome Research radiation hybrid map of the human genome.
(AC)n repeat analysis was carried out by using PCR primers to amplify the polymorphic region with the incorporation of a-^^P dCTP (Table 2.4). The products were electrophoresed in family groups on denaturing polyacrylamide gels and the alleles were detected on autoradiography film. In addition to microsatellites, a base change variant that resulted in the creation or absence of a restriction endonuclease site in the MSX2 gene was used as a marker. Alleles were scored for each family member, and the data was analysed by MLINK. For all families a penetrance level of 0.9 was used, and the disease frequency was entered as 1 x 10’^.
All of the pedigrees were too small to be able to test for significant linkage (that is, LOD > 3). Since only families 1 and 2 were classed as having the same syndrome, only their LOD scores could be added together; families 3 - 7 had to be considered separately. Exclusion rather than linkage of the genes was therefore tested for. Due to a penetrance level of 0.9 being applied, a LOD score of -oo at a recombination distance of 0 was never obtained when a crossover occurred (as would be expected if penetrance was 1). A LOD score of < -2 was considered to exclude the gene for the family. In some instances, families 1, 2 and 6 were too small to generate a LOD score of < -2 at 0.9 penetrance, despite crossovers. However, the chance of a crossover occurring between a locus and a closely linked marker in such a small number of meioses would be very low, and so exclusion on this basis was often concluded. Where exclusion could not be assumed, further screening analysis of the gene sequence was carried out.
3.3.1.2.1 FGFRl
FGFRl is known to be associated with Pfeiffer syndrome, and hence the gene was considered a candidate for the study.
Three markers mapping to the 8pl2 region of FGFRl were used. All were (AC)n repeats. F G F R lCA4 is located within the gene (Yu C-E, 1994), and the other two were both chosen from within the range identified from the Research Tools web page. D8S532 is located at an approximate maximum distance of 3.4 cM from FGFRl, and D8S1791 within 1.7 cM.
Table 3.6 LOD scores at recombination distances 0 - 0.4 for FGFRl
Family Marker locus Accession # 0 0.1 0.2 0.3 0.4
(GDB) (EMBL, GenBank) 1 D8S532 AFM081ydll -1.22 -0.56 -0.27 -0.11 -0.03 2 D8S532 AFM081ydll -0.44 -0.23 -0.11 -0.05 -0.01 3 -117 aff FGFRl FGFR1CA4 -5.14 0.33 0.41 0.31 0.15 -117 unaff -3.02 0.61 0.57 0.38 0.16 4 D8S1791 AFMc002zg5 -7.17 -1.00 -0.56 -0.29 -0.11 5 FGFRl FGFR1CA4 -6.21 -0.44 -0.19 -0.08 -0.02 6 D8S532 AFM081ydll -0.18 0.15 0.23 0.21 0.13 7 D8S532 AFM081ydll -6.14 -0.29 -0.12 -0.05 -0.01
The two families affected with CFND ( 1 and 2) had a minimum combined LOD score of -1.66 at 0 recombination distance. Crossovers had occurred in both families: in family 1 crossovers had occurred between affected individual III and the unaffected individuals TTT2 and III3, and in family 2 between affected individual II and unaffected 113 (Fig 3.38). However, in retrospect to the Xp22 linkage results of family 1 in section 3.4, it seems likely that unaffected ni3 actually carried the mutant gene. Therefore the aforementioned crossover involving III3 did not actually occur. However, FGFRl was still considered
excluded due to the presence of the two remaining crossovers and the evidence for linkage of CFND to Xp22. Families 3 and 5 both had LOD scores below -2 at 0 recombination distance (irrespective of whether individual 117 was assumed to be affected or unaffected) using the internal FGFRl marker. The gene was therefore also excluded in these families. Families 4 and 7 had low enough LOD scores at 0 and 0.1 recombination distances to give a LOD score below -2 at the respective maximum recombination distances of the markers used, and hence FGFRl was again excluded. Finally, although an insignificant LOD score was obtained for family 6 using a marker at a maximum distance of 3.4 cM from the gene, a crossover between affected individual 12 and unaffected III occurred, and so FGFRl was considered unlikely to be associated with the family.
3.3.1.2.2 FGFR2
A number of craniosynostosis syndromes are caused by FGFR2 mutations, such as Crouzon and Pfeiffer. Hence FGFR2 was a good candidate gene for the families.
FGFR2 maps to 10q25-26. Four different markers were used to obtain informative meioses in all pedigrees. Their positions relative to FGFR2 were obtained from an integrated genetic and physical map of chromosome 10 (http://pandora. cric.com/ htdocs/ sequences/chrlO-mapping/GenPhysMap.html). DIOS 1757 lies approximately 0.8 cM telomeric to FGFR2, D10S587 lies 3.7 cM centromeric, and D10S214 and D10S216 are both approximately 10 cM centromeric.
T able 3.7 LO D scores at recom bination distances 0 - 0.4 for FGFR2
Family Marker locus
(GDB) Accession # (EMBL, GenBank) 0 0.1 0.2 0.3 0.4 1 D10S214 AFM200yh6 -0.22 -0.06 -0.01 0.00 0.00 2 D10S216 AFM205zd8 -0.44 -0.23 -0.11 -0.05 -0.01 3 -117 aff D10S587 AFM296zg9 -11.89 -1.44 -0.82 -0.44 -0.18 - U7 unaff -10.23 -1.02 -0.54 -0.30 -0.15 4 D10S1757 AFM309yel -2.10 -1.14 -0.61 -0.29 -0.10 5 D10S1757 AFM309yel -5.97 -0.27 -0.11 -0.07 -0.05 6 D10S1757 AFM309yel -0.18 -0.02 0.02 0.03 0.02 7 D10S587 AFM296zg9 -5.80 -0.01 0.12 0.09 0.03
Pedigrees 1 and 2 both gave insignificant LOD scores for the different markers used; however, a crossover had occurred in each family: in family 1 a crossover had occurred between affected individual HI and unaffected EUS, and in family 2 between affected II and unaffected 113. However, as mentioned above in retrospect it seems likely that unaffected ni3 actually carried the mutant gene (section 3.4). Therefore the aforementioned crossover involving ni3 did not actually occur. However, FGFR2 was still considered excluded due to the presence of the remaining crossover in family 2 and the evidence for linkage of CFND to Xp22. Pedigree 6 gave an insignificant result, but a crossover between affected individual 112 and unaffected EH had occurred (Fig 3.39). FGFR2 was therefore considered unlikely to be the causative gene. Families 3, 4, 5 and 7 all gave LOD scores that would be below -2 at the recombination distance of the marker used, thereby excluding the gene in all cases.
3.3.1.2.3 FGFR3
Non-syndromic craniosynostosis phenotypes have been associated with a mutation in FGFR3. Crouzon syndrome with acanthosis nigricans is also caused by another mutation in the gene. Therefore FGFR3 was tested as a candidate gene.
Markers D4S2936, D4S43 (Tagle DA, 1992 ), D4S431 and D4S412 were used. They were positioned relative to a map of 4pl6.3 constructed during the localization of the achondroplasia gene (Francomano CA, 1994). Their approximate maximum distances from FGFR3 are 3.5 cM telomeric, 0.5 cM centromeric, 3.7 cM centromeric and 5.3 cM centromeric respectively.
Table 3.8 LOD scores at recombination distances 0 - 0.4 for FGFR3
Family Marker locus
(G DB) Accession # (EM B L, G enBank) 0 0.1 0.2 0.3 0.4 1 D4S412 AFM196xb6 -0.48 -0.24 -0.12 -0.05 -0.01 2 D4S412 AFM196xb6 - 6 . 2 2 -0.44 -0.19 -0.08 - 0 . 0 2 3 -117 aff D4S2936 AFMa275yg5 -10.28 -0.84 -0.41 -0.22 -0.11 -117 unaff -10.50 -1.01 -0.53 -0.29 - 0 . 1 2 4 D4S43 D4S43 -7.88 -0.69 -0.31 -0.12 -0.03 5 D4S2936 AFMa275yg5 0.56 0.43 0.29 0.15 0.04 6 D4S43 D4S43 0.56 0.47 0.38 0.27 0.14 7 D4S431 AFM262vg9 -6.06 -0.77 -0.35 -0.14 -0.03
Pedigrees 1 and 2 (Fig 3.40), affected with CFND, had a combined LOD score of -6.70 at 0 recombination distance, which would still give a value below -2 at the recombination distance of the marker used. Therefore FGFR3 was excluded. Similarly, the gene was excluded for families 3, 4 and 7.
Pedigree 6 gave an insignificant LOD score from two informative meioses with the marker D4S43, positioned centromeric to the gene. The informative meiosis did however follow the established phase. Two additional markers telomeric to the gene, D4S2936 and a marker at the PDEB locus (Weber B, 1993) which was approximately 1.5 cM from FGFR3, were used. Unfortunately the same pattern of uninformativeness was obtained. An affected family member had previously been screened for a mutation in the ser-pro dipeptide of the IgII-111 linker region of FGFR3 (section 3.2) where the mutation for
variable craniosynostosis, pro250arg, was located. No change was identified, and as confirmation the diagnostic Nci\ digest for the specific mutation was carried out. A PCR amplification, using a forward primer from exon 6 (FGFR3F) and a reverse primer from the 3’ end of exon 7 (FGFR3R1) (Table 2.3), yielded at 351 bp product including the pro250 site. Nci\ digestion of a normal allele would cleave the product into two fragments of 319 bp and 32 bp. However, the C-^G mutation creates an additional site, further digesting the 319 bp fragment into 168 bp and 151 bp. These bands would be identifiable by electrophoresis on a 1.5% nusieve/1.5% agarose gel. As expected, the test was negative for the mutation.
Similarly, family 5 gave an uninformative result. The above mutation tests were carried out on an affected family member, but again the tests were negative.
Due to the lack of a significant indication that FGFR3 was associated with the two families, and the exclusion of the only mutation published to date that would be expected with regard to the phenotypes of the two families, further analysis was not considered viable.
3.3.1.2.4 FGFR4
FGFR4 is the only member of the FGFR family in which mutations have not been identified. However, its homology to the other FGFRs rendered it a candidate gene in this study.
FGFR4 was tested using the 5q33-qter (GATA)n and (AC)n microsatellite markers D5S1456 and D5S429 (respectively). D4S1456 was chosen because it was located on the same contig (WC5.13) as AFMA109xa5, an STS identified as being close to FGFR4 by a radiation hybrid screen. The marker is approximately 4.5 cM from the gene. D5S429
was identifed from the region defined as containing the STS WI-6737, which was also identified as being close to FGFR4 by the radiation hybrid screen, on the Research Tools web page. It is located within an approximate distance of 2.2 cM from the gene.
Table 3.9 LOD scores at recombination distances 0 - 0.4 for FGFR4
Family Marker locus Accession # 0 0.1 0.2 0.3 0.4
(G DB) (EM B L, G enBank) 1 D5S1456 G A T A -llA ll -1.22 -0.56 -0.27 -0.11 -0.03 2 D5S1456 G A T A -llA ll -0.44 -0.23 -0.11 -0.05 -0.01 3 -117 aff D5S1456 G A T A -llA ll -8.80 -1.37 -0.62 -0.25 -0.07 -117 unaff -6.80 -1.16 -0.49 -0.19 -0.05 4 D5S429 AFM242xblO -6.10 -2.31 -1.30 -0.69 -0.28 5 D5S1456 G A T A -llA ll 0.56 0.43 0.29 0.15 0.04 6 D5S429 AFM242xblO -1.48 -0.78 -0.44 -0.23 -0.09 7 D5S1456 G A T A -llA ll -5.84 -0.23 -0.06 -0.01 0.00
Families 1 and 2 gave a combined LOD score of -1.66, and as crossovers had occurred in both pedigrees: in family 1 between affected individual III and unaffected individuals m i and ni3 (Fig 3.41), and in family 2 between affected individual II and unaffected 113. However, the aforementioned likelihood that unaffected m 3 in family 1 actually carried the mutant gene means that the aforementioned crossover involving m 3 did not actually occur. However, FGFR4 was still considered excluded due to the presence of the two remaining crossovers and the evidence for linkage of CFND to Xp22. Families 3, 4 and 7 all gave low enough LOD scores to be excluded at the recombination distances of the markers used. A crossover occurred in Family 6 between affected individual 12 and unaffected III, and so again the gene was excluded. Family 5 gave an insignificant LOD score due to uninformity of markers. Analysis was not taken further due to lack of significant evidence for the likelihood of an association with the disease.
Fig 3.38 Analysis o f family 2 for F G F R l, using marker D8S532. A crossover is seen between II and 113 11
O
12 1 112 113Fig 3.39 Analysis of family 6 for FGFR2, using marker DIOS 1757. A crossover is seen between 112 and III 1
a
Ô
1 1 1 1 1 2 12 113 nil 1112 1 2 3 4 5 6 7Fig 3.40 Analysis of family 2 for FGFR3, using marker D4S412. A crossover is seen between II and 112 11
O
12 111 112 113 1 2 3 4 5Fig 3.41 Analysis of family 1 for FGFR4, using marker D5S1456. Crossovers have occurred between III and both IIIl and III3
t
o 0
o
IIIl 1112 1113 1114
1 2 3 4 5
3.3.1.2.5 FGFI
FGFl is a high affinity ligand for all four FGFRs, and was therefore considered to be a candidate gene for the families.
FGFl is located on chromosome 5q31-33. Three markers were used in the analysis: one located in intronic sequence of the gene (Li X, 1992), a second, D5S2017, from within the range identified for the gene using the Research Tools web page, that maps within 2.7 cM of FG Fl, and D 5S178, positioned within 1.8 cM of the gene (Le Beau MM, 1993).
Table 3.10 LOD scores at recombination distances 0 - 0.4 for FGFl
Family Marker locus
(GDB) Accession # (EM B L, G enBank) 0 0.1 0.2 0.3 0.4 1 D5S178 D5S178 -0.22 -0.06 -0.01 0.00 0.00 2 FGFl FGFA.PCRl 0.56 0.43 0.29 0.15 0.04 3 -117 aff FGFl FGFA.PCRl -5.92 -0.48 -0.26 -0.14 -0.06 -117 unaff -6.11 -0.61 -0.33 -0.17 -0.07 4 D5S2017 AFMb074xgl 1.08 0.87 0.63 0.37 0.12 6 FGFl FGFA.PCRl -6.96 -1.16 -0.69 -0.39 -0.17 7 D5S2017 AFMb074xgl 0.56 0.43 0.29 0.15 0.04
Although family 2 was fully informative and no crossovers occurred within the pedigree, a crossover did occur in family 1 between affected individual III and unaffected m i,
totalling one out of five informative meioses in the CFND families. Therefore it was considered most likely that FGFl was not the disease causing gene. Families 3 and 6 (Fig 3.42) gave LOD scores below -2 at 0 recombination distance with the intragenic marker, and hence the gene was excluded. Despite using three additional markers (FGFA.PCRl, FGF1.PCR2 - GDB accession number 210943 - and D5S178) to obtain fully informative results for family 7, only two out of four meioses were informative. No crossovers were identified in the informative meioses. Hence an insignificant LOD score was obtained.
In family 4, three out of seven meioses were uninformative, although the four informative meioses did associate with the established phase. Further analysis of FGFl by SSCP was therefore undertaken for families 4 and 7.
FGFl comprises three coding exons, and intronic PCR primers were designed from nucleotide sequence available via ‘entrez’ on the internet (http://www3.ncbi.nlm.nih.gov/ htbin-post/Entrez/query?db=n&Dopt=g&form=6&form=6&uid=M23017) to amplify the exons and splice junctions within product fragments of 270bp, 265bp and 301 bp in length (Table 2.6). Two affected and two unaffected members of each family were amplified for each exon, and the products were labelled by the incorporation of a-^^P dCTP during the amplification. The samples were electrophoresed on non-denaturing gels with and without 5% glycerol, for 20 hours at 4°C.
The only difference between the band pattern of the affected and unaffected members of family 4 occurred in the PCR product encompassing exon 3. Two different bands (1 and 2) were present among the individuals: the two affected cases only had band 2, one unaffected case only had band 1, and the remaining unaffected case was heterozygous for the bands. The whole pedigree was then tested for the band pattern, but although all five affected members had band 2, individuals 112 and m3 were heterozygotes, as were unaffected merhbers 115, 116 and 117. In addition, ten normal controls were tested by running samples adjacently to the family on the SSCP gel analysis. One control was homozygous for the band 2 allele, five for band 1, and six were heterozygous. It was therefore assumed that the band shifts represented a polymorphism, and analysis was not continued. Unfortunately the variant did not segregate in a more informative pattern than the previously used markers. However, the absence of any detectable mutation in the family using two gel conditions implied that FGFl was unlikely to cause the disorder, and so the gene was excluded.
In family 7 the same polymorphic variant was detected in the exon 3 product, with both alleles occurring in affected and unaffected members of the family. No band differences
were observed in the exon 2 product. However, a different band pattern was observed in the two unaffected (and unrelated) members of the family to the affected members. The two remaining affected members were subsequently tested, and were shown to share the same pattern as the other affected cases. However, 22 unrelated normal controls also shared the affected pattern, and it was therefore concluded that the unaffected family members carried a less common variant allele. Hence, FGFl was again excluded as the causative gene.
3.3.1.2.6 FGF2
FGF2 is a ligand for the Hie isoforms of FGFRl, -2 and -3 and was therefore considered as a candidate gene.
An (AC)n microsatellite repeat, D4S430, was chosen from the range given for the location of FGF2 on chromosome 4q25-q27 on the Research Tools web page. It is positioned at an approximate maximum distance of 3.8 cM from the gene.
Table 3.11 LOD scores at recombination distances 0 - 0.4 for FGF2
Family Marker locus Accession # 0 0.1 0.2 0.3 0.4
(G DB) (EM BL, GenBank)
1 .D4S430 AFM260za5 -0.22 -0.06 -0.01 0.00 0.00
6 D4S430 AFM260za5 0.56 0.47 0.38 0.27 0.14
A crossover in family 1 between affected individual III and unaffected individual ni3 lead to FGF2 being considered excluded from causing the CFND phenotype of the family (Fig 3.43). However, in retrospect of the CFND linkage data in section 3.4, where unaffected III3 was shown most likely to carry the mutant gene, this crossover did not
occur, and FGF2 could not be excluded on this basis. However, since the results of section 3.4 indicate that family 1 links to Xp22, then FGF2 can be excluded anyway. Family 61 showed two out of a possible three informative meioses , but resulted in an insignificant LOD score. The data was not considered significant enough for further analysis of the gene.
3.3.1.2.7 FGF3 andFGF4
FGF3 and -4 are located very close to each other on chromosome llq l3 . FGF4 was considered to be a good candidate since it is a high affinity ligand to the rUc isoforms of FGFRl, -2 and -3. FGF3 binds to FGFRl and -2 m b isoforms, and was also tested as a